[Paper Review] A General Framework for Performance Analysis of Spatial Modulation over Correlated Fading Channels

Mutlu Koca, Hikmet Sari|arXiv (Cornell University)|Sep 26, 2011

Advanced Wireless Communication Technologies15 references3 citations

TL;DR

This paper proposes a general analytical framework for performance evaluation of spatial modulation (SM) over correlated Rayleigh and Rician fading channels using proper complex Gaussian modeling of the error metric. By leveraging the multivariate distribution of the fading channel vector, it derives tight upper bounds on the pairwise error probability (PEP) for arbitrary numbers of transmit and receive antennas and general linear constellations, with exact closed-form expressions for SSK as a special case.

ABSTRACT

We present a general method for the error analysis of spatial modulation (SM) systems over correlated and uncorrelated Rayleigh and Rician fading channels. The proposed method, making use of the properties of proper complex random variables and vectors, provides an exact upper bound for the class of fading channels considered for any number of transmit and receive antennas and for a wide family of linear modulation alphabets. Theoretical derivations are validated via simulation results.

Motivation & Objective

• Address the lack of a general performance analysis framework for spatial modulation (SM) over correlated fading channels with arbitrary antenna configurations.
• Overcome the challenge of deriving closed-form pairwise error probability (PEP) expressions due to the complex, correlated nature of the SM error metric.
• Extend existing SM performance bounds—previously limited to real constellations or uncorrelated channels—to complex constellations and correlated Rayleigh/Rician fading.
• Provide a unified analytical approach applicable to both SM and its special case, space-shift keying (SSK), with exact upper bounds.
• Enable accurate performance evaluation for practical MIMO systems with high spectral efficiency and realistic channel correlation models.

Proposed method

• Model the SM error metric using a proper complex Gaussian vector derived from the channel fading coefficients and constellation symbols.
• Utilize the joint multivariate Gaussian distribution of the complex fading vector to compute the PEP upper bound without relying on envelope-based PDF or MGF methods.
• Apply the moment generating function (MGF) approach from [15] to derive exact upper bounds on the average pairwise error probability (APEP) for correlated fading channels.
• Formulate the upper bound using the determinant of a matrix involving the correlation matrices and signal-to-noise ratio (SNR), enabling closed-form computation for uncorrelated and correlated cases.
• Extend the framework to SSK by treating it as SM with a single-constellation-point alphabet (M=1), allowing reuse of the same analytical structure.
• Validate the analytical bounds via Monte Carlo simulations across various fading conditions, including uncorrelated and correlated Rician fading with different correlation parameters.

Experimental results

Research questions

• RQ1 Can a general, closed-form upper bound for the pairwise error probability (PEP) of spatial modulation be derived over correlated Rayleigh and Rician fading channels?
• RQ2 How can the performance of SM be analyzed when the error metric involves a complex, weighted sum of correlated fading vectors with complex constellations?
• RQ3 To what extent can the performance of space-shift keying (SSK) be analyzed using the same framework as SM, particularly in correlated fading environments?
• RQ4 How tight are the derived upper bounds compared to simulation results across different spectral efficiencies and correlation models?
• RQ5 Can the proposed method be applied to arbitrary linear modulation formats (e.g., QAM) and arbitrary numbers of transmit and receive antennas?

Key findings

• The proposed framework provides an exact upper bound for the average pairwise error probability (APEP) of SM over correlated Rayleigh and Rician fading channels with arbitrary numbers of transmit and receive antennas.
• For uncorrelated Rician fading, the APEP upper bound is derived in closed form using the determinant of a matrix involving the correlation and SNR parameters, enabling efficient performance evaluation.
• In correlated fading scenarios, the upper bound remains tight across all tested conditions, including exponential and Kronecker correlation models with varying correlation levels (e.g., γt=γr=0.8, γt=0.9, γr=0.1).
• The framework successfully generalizes to SSK by treating it as SM with a constellation size of 1, yielding a consistent and analytically tractable upper bound that matches simulation results.
• Simulation results show a tight match between the theoretical upper bounds and actual bit error rate (BER) performance down to 1×10⁻⁶ BER across all spectral efficiencies (R=3 to 7 b/s/Hz) and fading models.
• The method is robust across different modulation formats, including QPSK, 8-ary, 16-ary, and 32-ary rectangular QAM, confirming its generality and applicability to practical MIMO systems.